Open PHD

[Jabroni]

I've been playing around with open phd, trying to add a function to name the stars that the guidescope is showing.

Here's what I'm doing:
- using PHD's "Auto select star" algorithm to find the brightest stars in the image.
- work out the distance and angle between the brightest stars.

http://www.iceinspace.com.au/forum/attachment.php?attachmentid=109048&stc=1&d=1329013554

- Loading the Tycho2 catalog, and moving the stars based on their yearly movement.
- Finding the best match in the catalog for the identified angles and distances.

Here are the results:
(Orion SSAG screen shots used through the phd simulator camera).

NGC 3532
http://www.iceinspace.com.au/forum/attachment.php?attachmentid=109046&stc=1&d=1329012504

NGC 3372
http://www.iceinspace.com.au/forum/attachment.php?attachmentid=109047&stc=1&d=1329012504

I still have a few problems. The two red stars in the first image are critical - the distance and angle of these two are used as a reference for all other stars. If an image has a really bright star, sometimes these two end up on the one star. I'd like to add a method to manually remove and add "tagged" stars.

If there's a planet, or other solar object tagged as a star, it would mess things up too.

I still need to add a settings window to set the searched field of view, error tolerances, number of stars, threads used for calculation, etc. Once this is done, I might try incorporating it back into the open phd project.

What would be nice is a Stellarium plugin, so when you click on a star in PHD, it selects it in Stellarium.

Anyway, let me know what you think...

[alistairsam]

That sounds pretty cool Jason.
Are you modifying the open phd files or do you have an add-on to recognize and label stars.
is this called plate solving?

[Jabroni]

I'm modifying phd - it doesn't have support for standalone plugins.
All my code is in it's own class(es), and I've only modified what was neccesary for menu entries, etc.

I think it's called blind plate solving - solving without any hints.

It's fairly fast with a narrow FOV (4 degrees), and only stars with magnitudes brighter than 8 or 9. Both examples solved in about a second.

[allan gould]

That's clever!
Be interested to see how it turns out.

[Jabroni]

Here's a sneak peak. It's the open linux version compiled in windows, so it doesn't have all the good drivers.

http://jabroni.org/NameStars.zip

- Connect to the simulator camera
- Loop exposures and it will random between 6 images (From my starshoot autoguider).
- Stop on an image

Everything new is at the bottom of the tools menu:
- Select stars. If it miss-identifies anything, Shift-Click to remove the incorrect selected star.
- Name Stars.

There's a configuration window now (the settings aren't persistent when you quit yet).

When you select any star from the camera image it tries to name it (using the full catalog). If the search pattern matched correctly, it should be able to name almost every star in the frame.

"Catalog Magnitude Threshold" has a huge impact on search speed. It only needs to be high enough to cover most of the stars in the search pattern. As it approches around 11, an order of magnitude can mean around another million stars to process...

The catalog only loads on the first naming cycle, or when you change the above value.

The number of search threads defaults to half your number of cores/threaded cores.